Geth rakudo: MasterDuke17++ created pull request #1080:
Test instantiating non-class has its name in error		 00:47
rakudo/nom: 4666505570 | MasterDuke17++ | t/05-messages/01-errors.t
Test instantiating non-class has its name in error

For RT #122232		 00:57
rakudo/nom: 6435933a1d | (Zoffix Znet)++ (committed using GitHub Web editor) | t/05-messages/01-errors.t
Merge pull request #1080 from MasterDuke17/tests_for_RT122232

Test instantiating non-class has its name in error
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=122232
MasterDuke_ thanks, gonna resolve that ticket now 00:58
j: my @a; dd [>>+<<] @a 01:25
camelia ( no output )
MasterDuke_ m: my @a; dd [>>+<<] @a 01:26
camelia Too few positionals passed; expected 2 arguments but got 0
in block <unit> at <tmp> line 1
Geth roast: 1244df8c4a | MasterDuke17++ | S03-metaops/reduce.t
Test reduce metaop of hyper metaop with ...

0 and 1 element arrays.
RT #122475		 01:47
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=122475
MasterDuke_ j: my @a = $(1, 2, 3); dd [>>+<<] @a;
m: say &OUR::bar() 02:44
camelia Cannot find method 'Any' on object of type List
in block <unit> at <tmp> line 1
MasterDuke_ m: my %h{Any} of Str; %h<a>=2 03:01
camelia Type check failed in binding to parameter 'assignval'; expected Str but got Int (2)
in block <unit> at <tmp> line 1
MasterDuke_ m: my %h of Str; %h<a>=2 03:02
camelia Type check failed in assignment to %h; expected Str but got Int (2)
in block <unit> at <tmp> line 1
MasterDuke_ that first error is LTA. what is `parameter 'assignval'`? 03:03
bisect: my %h{Any} of Str; %h<a>=2
bisectable6 MasterDuke_, Bisecting by output (old=2015.12 new=6435933) because on both starting points the exit code is 1
MasterDuke_, bisect log: gist.github.com/4403a1a13e795dea89...a2ae8cef07
MasterDuke_, (2016-02-18) github.com/rakudo/rakudo/commit/99...4283b48d0a
MasterDuke_ c: all my %h{Any} of Str; %h<a>=2 03:04
committable6 MasterDuke_, gist.github.com/1d16e219f69480339b...30bc4eb987 03:05
MasterDuke_ bisect: old=2015.10 new=2015.11 my %h{Any} of Str; %h<a>=2 03:06
bisectable6 MasterDuke_, Bisecting by output (old=2015.10 new=2015.11) because on both starting points the exit code is 1
MasterDuke_, bisect log: gist.github.com/4b5c9a27278501e664...3037ee8b2c
MasterDuke_, (2015-11-18) github.com/rakudo/rakudo/commit/06...7ee518b737
lizmat Files=1192, Tests=56909, 202 wallclock secs (12.06 usr 4.60 sys + 1198.63 cusr 116.24 csys = 1331.53 CPU) 07:13
commute to PTS@Lyon & 07:51
Zoffix m: my $i = -0x8000000000000000; say ($i +> 32) 10:18
camelia -1
Zoffix "a < 0 && b > 31 ?? -1 # temp fix for RT#126942, remove if fixed otherwise" 10:19
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126942
Zoffix m: dd nqp::bitshiftr_I(-0x8000000000000000, 32, Int) 10:20
camelia 5===SORRY!5=== Error while compiling <tmp>
Could not find nqp::bitshiftr_I, did you forget 'use nqp;' ?
at <tmp>:1
------> 3itshiftr_I(-0x8000000000000000, 32, Int)7ā5<EOL>
Zoffix m: use nqp; dd nqp::bitshiftr_I(-0x8000000000000000, 32, Int)
camelia -2147483648
Zoffix m: use nqp; dd nqp::bitshiftr_I(-0x8000000000000000, 64, Int)
camelia -1
Zoffix m: use nqp; dd nqp::bitshiftr_I(-0x8000000000000000, 40, Int)
camelia -8388608
Zoffix m: use nqp; dd nqp::bitshiftr_I(-0x8000000000000000, 70, Int)
camelia -1
Zoffix m: use nqp; dd nqp::bitshiftr_I(-0x8000000000000000, 63, Int)
camelia -1
Zoffix m: use nqp; dd nqp::bitshiftr_I(-0x8000000000000000, 61, Int)
camelia -4
Zoffix So looks like it works, it's the kludge that actually is problematic 10:21
will nix it in couple of hours to fix 131278
m: use nqp; dd nqp::bitshiftr_I(-0x8000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000, 700, Int) 10:39
camelia -2048
Zoffix And bigint works too so I guess that NIY can go too 10:40
And removing all that stuff makes the op 55% faster too :) 10:45
ZofBot: this is the best bug fix ever!
ZofBot Zoffix, See S06 for more about binding
[TuxCM] running timing from train 10:54
This is Rakudo version 2017.04.3-210-g6435933a1 built on MoarVM version 2017.04-57-g8d8a09b9 11:02
csv-ip5xs 2.931
test 12.573
test-t 4.248 - 4.259
csv-parser 12.614
Zoffix Timing from the train? Cue Prof. Einstein :) 12:10
Not fair! You made it run slower by moving at some velocity :) 12:11
timotimo timings have to be performed at the same distance from sea level and far away from any big mountains 12:14
moritz as long as you measure the time within your own frame of reference, all is good :-) 12:24
Zoffix :)
timotimo you must be careful when the train is going through curves, though
moritz not really 12:25
timotimo sideways acceleration will throw you out of an equivalent inertial reference frame to outside of curves!
moritz you only have to be careful if the system you're timing is in a different frame of reference than your clock
timotimo what if an ICE drives though a curve of radius 100 meters at 300 km/h?
moritz then you have other things to worry about :-) 12:26
timotimo :D
moritz like, picking up all your body parts
timotimo wolframalpha says if the mass is 1kg it'll just be 69.44 newtons 12:27
so you can just multiply that with your body weight i guess?
Zoffix That's 70g. very fatal 12:28
moritz Zoffix: more like 7g
Zoffix: 1g is 9.8m/ss
Zoffix Ah right
timotimo hey it's 2200x times the "thust of the NSTAR ion engine tested on NASA's space probe"
Zoffix heh
moritz so, something that you can survive if you're strapped tightly in a cushioned seat
but sitting in a regular ICE seat, it'll smack you against the nearest wall, and if you're too far away from that wall, you're... very much out of luck 12:29
timotimo oh, you were not refering to grams, you were refering to "times earth's average gravitational acceleration" 12:32
i thought "man, 70g, yeah, that's so very fatal!"
timotimo looks at 100g of chocolate next to him
Zoffix ZOFFLOP: t/spec/S11-modules/nested.t 12:42
brrt there are plenty of substances of which 70g is fatal 12:44
timotimo finds out that --hash-style=both is not worth a lot for libmoar 12:45
to be exact: -Wl,--hash-style=both
MasterDuke_ c: 06729d9~1,06729d9 my %h{Any} of Str; %h<a>=2 12:48
committable6 MasterDuke_, gist.github.com/acaa9422e1c6af7876...6727bbb3bb
MasterDuke_ jnthn: any idea why github.com/rakudo/rakudo/commit/06...7ee518b737 caused the error messages from ^^^ to lose the name of the variable? 12:49
timotimo oh, it makes zero difference when you're using lazy binding, which is the default 12:50
Zoffix cpan@perlbuild4~/CPANPRC/rakudo (nom)$ ./perl6 -e 'say -15 +> 50' 12:53
0
cpan@perlbuild4~/CPANPRC/rakudo (nom)$ ./perl6 -e 'say -15**1 +> 50'
-1
dafuq?
m: use nqp; say nqp::bitshiftr_I(nqp::decont(-15), nqp::unbox_i(50), Int); say nqp::bitshiftr_I(nqp::decont(-15**1), nqp::unbox_i(50), Int); 12:56
camelia 0
-1
Zoffix And I wrote 176 tests for this bug and all of them are green because I used powers to make the numbers to shift -_- 12:59
nine Yeah ye olde truth that tests can never show the absence of bugs, only the presence. 13:00
Zoffix :)
looks to be MoarVM-only 13:01
Zoffix digs
Well, this is the easiest fix ever it seems 13:09
timotimo oh? 13:10
interested to hear what the problem behind this bug was
Zoffix Yeah, it just sets it to 0 but needs to set to -1
timotimo hah
Zoffix :)
timotimo so not even calculating it wrong
Zoffix nope
timotimo just a wrong special case?
Zoffix Yeah
timotimo and how does it do it wrong in one case and right in the other? 13:11
Zoffix Dunno, I wonder if MVM_BIGINT_IS_BIG returns true for -15**1 13:12
Well, for that case I mean
Zoffix tests
yup, it goes through that path 13:15
So, another bug in MVM_BIGINT_IS_BIG ?
timotimo well, it's supposed to give the same results in both paths 13:16
Zoffix #define MVM_BIGINT_IS_BIG(body) ((body)->u.smallint.flag != 0xFFFFFFFF)
And I guess power op sets that flag?
timotimo probably, if the result is small
Zoffix Yeah, it gives same results now that I fixed the stuff
timotimo i wonder why we don't use smallbigint for -15 there even though we could 13:17
perhaps we do that for no literals at all
that could be a nice little improvement to fix that
Zoffix mhm
PR sent: github.com/MoarVM/MoarVM/pull/593 and I got a Rakudo patch to proper-fix RT#131278 and RT#126942 once that PR is merged 13:34
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131278
Link: rt.perl.org/rt3/Public/Bug/Display...?id=126942
Zoffix nine: are you too at the Perl Toolchain thing? 13:42
Zoffix & 13:49
nine Zoffix: yep :)
Zoffix Ah cool :)
timotimo Zoffix: could it be you've got the fakely-big ints part reversed? is it really -15**1 that doesn't know to set the "not actually big" flag? 14:01
i thought it was the other way around, but i didn't actually check
timotimo re-reads
Zoffix timotimo: MVM_BIGINT_IS_BIG returns true for -15**1 14:02
timotimo ah, ok, so BIGINT_IS_BIG rightly returns false for -15
that's a relief at least
but we'll also want to have a piece of code in there that can put the result of ** into a not-actually-big bigint
Zoffix s: &infix:<**>, \(-15, 1) 14:03
SourceBaby Zoffix, Sauce is at github.com/rakudo/rakudo/blob/6435...nt.pm#L302
timotimo nqp-level code has no way to set/clear/read that flag
oh, actually
i think we have a way to read it
m: use nqp; say nqp::isbig_I(-15); say nqp::isbig_I(-15**1)
camelia 0
0
timotimo maybe not the same thing 14:04
Zoffix m: use nqp; say nqp::bitshiftr_I(nqp::pow_I(-15, 1, Num, Int), nqp::unbox_i(50), Int);
camelia -1
Zoffix m: use nqp; say nqp::bitshiftr_I(nqp::pow_i(-15, 1), nqp::unbox_i(50), Int);
camelia 0
Zoffix wants a SourceBaby for Moar 14:06
Give it an NQP op and it tells you where moarvm source for it is at 14:07
nine What are repossession conflicts again?
timotimo that'd be neat, but don't forget there's also "virtual" ops you'd need the right repr for
nine: when a compilation unit does some kind of change to some sort of object, it'll claim the source object for itself and make the modifications
that way when a different compunit tries to change the same object in a different way, that's the kind of conflict you get 14:08
nine timotimo: mind if I paste that into a comment? 14:09
timotimo i'm not sure if it's accurate enough
nine It makes sense of me and for sure is better than nothing :0
:)
timotimo if it's a comment in the code that can be edited later, that's fine 14:10
a commit message, not so much :P
nine Yeah, I'll just add it to the comment above resolve_repossession_conflicts
timotimo OK, cool 14:11
Zoffix huh, interesting. use nqp; say nqp::pow_I(-15, 1, Num, Int) apparently executes MVM_bigint_pow twice if my debug print statements are to be believed :S 14:26
^ probably just another `pow` done during parsing or whatever 16:18
Geth rakudo/fix-right-shift: ef29bb9f41 | (Zoffix Znet)++ | src/core/Int.pm
Fix right shift op; make it 55% faster

  - Fix underlying MoarVM problem:
   github.com/MoarVM/MoarVM/pull/593
  - Remove the kludge and bigI check, making the op 55% faster
  - Fixes RT#126942: rt.perl.org/Ticket/Display.html?id=126942
  - Fixes RT#13278: rt.perl.org/Ticket/Display.html?id=131278		 16:21
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126942
roast/fix-right-shift: b7fac2678c | (Zoffix Znet)++ | S03-operators/bit.t
Test right bit shift for negative numbers

RT#126942: rt.perl.org/Ticket/Display.html?id=126942 RT#13278: rt.perl.org/Ticket/Display.html?id=131278		 16:22
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126942
rakudo/nom: 65c0786d3d | (Zoffix Znet)++ | t/05-messages/01-errors.t
Improve test

  - Test for larger string in error message
  - Addresses github.com/rakudo/rakudo/pull/1080...-300454314		 16:31
Zoffix Sheesh... All the names are taken! 17:00
I even went as whacky as "SpaceBear web framework"... And lo and behold: github.com/breuleux/spacebear 17:01
Not only is the name taken but it's also a web framework -_-
timotimo >_< 17:05
Zoffix goes with `Tardigrade` and using `TG::` namespace
[Coke] d 17:52
Zoffix So OSCON is happening? Do we have any Perl 6 stuff at it this year? 18:08
Zoffix glances at DrForr
Nothing really comes up when searching for perl so I guess not :) 18:18
timotimo yeah, drforr did a perl6 thing
i saw it on his twitter
Zoffix This year?
Aye
Sweet 18:19
Now I see it: conferences.oreilly.com/oscon/osco...tail/56965
Ah right, there's a link on TPF, but I got confused with "2016" mentioned in URL and thought it was a dated post news.perlfoundation.org/2017/05/per...ights.html 18:24
cool I'm mentioned there :o
:)
perlpilot Zoffix++ (tardigrade) 18:27
Zoffix ZOFFLOP: t/spec/S17-supply/interval.t 19:29
man, why do auto-checkouts of nqp/roast always have some sort of issues with remotes -_- 19:33
Permission denied (publickey).
fatal: Could not read from remote repository.
timotimo Zoffix: probably difference between https:// and ssh:// adn such 20:29
you can set a GIT_PROTOCOL (or something) env var before Configure.pl does its checkouts
Zoffix ah cool 20:30
samcv my new GraphemeBreaktest.t breaks when it passes through the fudger 20:35
looks like it goes throug hthe fudger because S15-nfg/GraphemeBreakTest.t # moar
in spectest.data?
how can i get it to run as a plain .t file? 20:36
timotimo without fudging, eh? 20:49
sometimes the fudger needs a bit of assistance to make fudging commands work, but i don't think it can stumble when there aren't any in the file
samcv i mean. shouldn't it just do nothing? 21:00
if there's no #?rakudo.moar lines or whatever
i mean there's plenty of scripts that run as testname.t though some run as testname.t.moarvm
timotimo yeah, it should do nothing 21:01
it shouldn't even make a .moarvm.t file
samcv yeah :\
timotimo you could diff the .t against the .moar.t or what it is
dinner time! 21:02
samcv you mean t.moar?
it's bytecode it runs
t/spec/S15-nfg/GraphemeBreakTest.rakudo.moar
oh it is an actual script my bad
but it removed a ton of thigngs... 21:03
Zoffix could be due to fudger not doing anything to handle unicode properly? 21:11
samcv argh ok i got it not to mangle most of it 21:12
by using #?DOES num
at the very top. but now it like... oved some brackets around
though everything else is ok
there's no unicode in the file itself
previously it ate up like 2/3 of the file 21:13
Zoffix There's whole bunch of unicode in github.com/perl6/roast/blob/master...reakTest.t 21:14
samcv now gist.github.com/samcv/c415977939cd...e1a79074ef
not in the new version
Zoffix samcv: ah. Does it still have fudges? The symptoms you describe sound like a fudged loop where you wrote, say `#?rakudo.moar 10 todo` when it should be 1 21:15
What's the full content of the file?
Geth nqp: 9763728215 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
Bump Moar

Moar VM bump brought: github.com/MoarVM/MoarVM/compare/2...2-g362277b
362277b Better version of patch to fix right shifting negative smallints f8bc328 Merge pull request #593 from zoffixznet/fix-bitshiftr_I 9d7bee4 Use -1 instead of 0 when long right-shifting negative smallints 957759c Merge pull request #591 from samcv/master 3e4ed6b Add can_fit_into_8bit funct, put logic used many places into one funct 8d8a09b telemeh can apparently bring in a stray dependency		 21:16
Ā¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...2-g362277b
rakudo/nom: 66e8e72c02 | (Zoffix Znet)++ | tools/build/NQP_REVISION
Bump NQP

NQP bump brought:
  github.com/perl6/nqp/compare/2017....6-g9763728
9763728 Bump Moar 674cb40 [js] Handle all the modes that can be passed to nqp::open 66eae59 [js] Remove debugging leftover ... (10 more lines)
Ā¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....6-g9763728
rakudo: zoffixznet++ created pull request #1081:
Fix right shift op; make it 55% faster
rakudo/nom: ef29bb9f41 | (Zoffix Znet)++ | src/core/Int.pm
Fix right shift op; make it 55% faster

  - Fix underlying MoarVM problem:
   github.com/MoarVM/MoarVM/pull/593
  - Remove the kludge and bigI check, making the op 55% faster
  - Fixes RT#126942: rt.perl.org/Ticket/Display.html?id=126942
  - Fixes RT#13278: rt.perl.org/Ticket/Display.html?id=131278
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126942
rakudo/nom: d6798bfe60 | (Zoffix Znet)++ (committed using GitHub Web editor) | src/core/Int.pm
Merge pull request #1081 from rakudo/fix-right-shift

Fix right shift op; make it 55% faster
roast: zoffixznet++ created pull request #266:
Test right bit shift for negative numbers
roast: 02d1f9d0f6 | (Zoffix Znet)++ (committed using GitHub Web editor) | S03-operators/bit.t
Test right bit shift for negative numbers (#266)

RT#126942: rt.perl.org/Ticket/Display.html?id=126942 RT#13278: rt.perl.org/Ticket/Display.html?id=131278
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126942
Zoffix $ ./perl6 t/spec/S03-operators/bit.t 21:24
Segmentation fault
:(
after making this change: gist.github.com/zoffixznet/2d017e9...bbafd89e52 21:25
jnthn Zoffix: With any local MoarVM patches? 21:26
Ah, I see recent patches went in on that
Zoffix oh wait, maybe not that change but the rakudo PR merge 21:28
Zoffix resets and tests
samcv Zoffix, you pulled in my changes right? 21:30
Zoffix samcv: yeah. Problem is, rakudo's tests didn't exercise your changes at all, because of a kludge 21:31
samcv those two tests you gave me worked
hm
the ones you said didn't work before
<Zoffix> And in rakudo `./perl6 -e '2 +> 100'` should give 0 and `./perl6 -e '-2 +> 100'` needs to give -1
these worked with the patch. but you say the function wasn't being used?
Zoffix Zoffix ā”‚ Oh, they might not give that without my rakudo patch, but, this will: 21:32
Zoffix ā”‚ m: use nqp; say nqp::bitshiftr_I(-2, 50, Int); # wrong; should give -1
Yeah, that one crashes
samcv works for me
with my latest pushed in moarvm commit 15 mins ago 21:33
gives me -1
Zoffix $ ./perl6 -e 'say $*VM.version'
samcv i have 2017.04.3-207-g6bb1b5b40
Zoffix v2017.03.232.gf.8.bc.328
samcv that's yours? 21:34
i have 2017.04-62-g362277b7
Zoffix Yeah, I don't get it. nqp/tools/build/MOAR... gives 2017.04-62 21:35
Well, as good a time as any to blow rakudo dir
samcv just update moarvm 21:36
i didn't bump nqp's moar revision
did you assume i bumped it somehow?
Zoffix I bumped it 21:37
samcv oh ok
ok i see that now
Zoffix Yeah, it was just a busted build. I guess it saw my moarvm already at specified version and did not rebuild it 21:41
timotimo yeah, i always go into the individual folders 21:42
samcv not fun. have to figure out why the fudger does this: gist.github.com/samcv/c415977939cd...e1a79074ef 21:43
to my file
just moves around my brackets 'for fun'
timotimo it must find the # before that somehow interesting
can you give more context above?
samcv this is it github.com/samcv/Unicode-Grant/blo...Grammar.p6 but saved as t/spec/S15/GraphemeClusterBreakTest.t 21:44
and so if you try and make t/spec... for that one. it deletes half of the file
so now i put #?DOES 2411 on line 7 21:45
and now it does what i showed you above instead of totally destroying everything
timotimo the only thing i can think of is the ? at the end of one of the comments
but unlikely
can you throw out the comment block (or all of them) and see if that changes things
samcv this is it before adding the #?DOES line: gist.github.com/4f6da1fd8353ac59cd...1ef31a739d 21:46
hmm k
timotimo what, it throws out that whole stuff completely? 21:47
samcv yes
it's awesome
timotimo, taking out the comments changes nothing 21:51
timotimo buh
samcv timotimo, moved things around and then it tries removing the whole file again too. like it's really odd. i just want it to not mess with the file 22:01
timotimo hm, is the fudger written in perl 5 or 6?
samcv perl 5 22:02
timotimo it's 5, yeah
i can't 5
Zoffix samcv: it's the shebang
samcv oh
timotimo seriously? the shebang makes it b0rk? 22:03
samcv yay
now it works... goddamit
how do we fix this
Zoffix samcv: amend this regex to ignore shebangs would be my guess: github.com/perl6/roast/blob/master/fudge#L206 22:04
timotimo put a check in there for the first line only?
samcv hmm thx
Geth roast: 2b0c18926d | (Zoffix Znet)++ | S03-operators/bit.t
Add more +> tests
 22:12
rakudo/nom: 6409ee582e | (Zoffix Znet)++ | src/core/Int.pm
Remove big int check for +<

It's no longer NYI		 22:13
roast: samcv++ created pull request #267:
[unicode-grant] Commit New version of GraphemeBreakTest.t		 22:34
samcv there we go :) finally ready to go 22:35
wow the diff sucks. github didn't realize the file was totally deleted 22:36
so viewing the diff looks terrible
timotimo heh. 22:38
samcv gonna split up the commits 22:47
ok much better now
timotimo, wanna quick review so i can merge? github.com/perl6/roast/pull/267 22:48
hopefully you approve of the folder layout for 3rdparty
and the layout of Unicode/9.0.0/ucd/auxilary.... # this is the same layout that's on unicode ftp 22:49
so there won't be any conflicts if we add things in the future
s/if/when
wait. nvm github is still dumb. it still shows it as a diff even tho i seperated the deletion commit 22:51
timotimo right, it's a 100% add diff 22:59
samcv yep
you can just go to the commit i add it and it shows up fine. otherwise it's garbage 23:00
timotimo right 23:01
i'm having trouble concentrating right now 23:03
samcv ok :) 23:10

Powered by Cro and the RakuĀ® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!